摘要 |
PURPOSE:To generate a conversational sentence from the voice of a speaker without using a language processor which uses complex algorithm. CONSTITUTION:When the speaker voices a word, a phoneme series generation part 1 outputs a phoneme series Nc from the speech and a word analytic part 2 after generating a re-edited phoneme series Ld by connecting the phoneme Nc to the tail of a phoneme series Lc whose word is indeterminate generates a word string Ln by using a word dictionary and outputs it to a semantic analytic part 3. The semantic analytic part 3 after generating a word string Ly for analysis by connecting the word string Ln to the tail of a word string Lw whose meaning is indeterminate generates syntax whose meaning is analyzed by using a semantic data base 4 and outputs it to a semantical form generation part 5. The semantical form generation part 5 generates syntax having the consistency of the front-rear relation of context from the syntax. A conversational sentence generation part 6 generates the conversational sentence from the syntax received from the semantical form generation part 5. |